So the MQA endpoint is not the DSP speaker. Would the MQA endpoint need to know which speaker (5200SE, 7200SE, 8000SE) is playing the audio in order to correctly correct (!)the A2D errors?
I believe so, and presumably this would be achieved through MHR and/or SL (and legacy?) comms.
I believe that the speaker is able communicate over SL what its capabilities are; for the oldest SL speakers which do not do this, absence of capability information can be used to infer capability!
I imagine, but don't know, that all legacy (pre-SL) DSP speakers will be lumped in the same boat.